xml_col_finder 0.1.0 → 0.2.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7ad99ebebeb440fb2a0229c49d2ab7e993ea086375bb4cf48264b07897e4c9cc
-  data.tar.gz: f4d6ba7b029cbfff03f147ed4c0da760651f34fa5fca46d675435ed1d78226fd
+  metadata.gz: b98fdd5a28880c8560256ae8a01f5e0a64a3067f2c0974f5c0e668243ce7ad92
+  data.tar.gz: c1177f458579fb3fd7a9d975512d14f5b0e2395ea4cbaa21dc40253b564b78f3
 SHA512:
-  metadata.gz: d7daa28d76d6e6f4d1ffc57041cbd4b24b96352bd10b85fe80d65eaa071e8c7e97e77b0ecb877f89e67445ce93aa138a652175da2509a0e853ff13839bfbb48d
-  data.tar.gz: c689924e6f815fa8f4398fb54b8142b4d1ad135f391d0c2836b4a93395c8a58e31b70f42ae05bf8c1b977fde6cc3d8a1c041ce29a3c86f0ef400720f6684f212
+  metadata.gz: 73daefd3b3bb3873e21848aad39fd7d9bb570229c8369bb71920bce81474d312b6eb83a9647cf03d66bc4b095788a72dee9925a68d8de8e78554f82155afbdea
+  data.tar.gz: be011c43afe08767e860c436024c65d8b86e507f610ffd7caad367f034cd76dfbfaf5c39882eef9ee6524cc2f90d1de9c564db297f4899df01f3e40e42840885

data/lib/xml_col_finder.rb

@@ -24,11 +24,80 @@ class XMLColFinder
 
     end
 
+    #@to_a = a
     h = group_by_xpath(a)
     @to_a = truncate_xpath(h).flatten(1)
 
   end
 
+  def to_code(nametip: true)
+
+    @nametip = nametip
+    @tags = {}
+
+    xpath, remaining = @to_a
+
+    eid = getid(xpath)
+    linex = formatline('doc', eid, xpath)
+    a = scan(remaining, eid)
+
+    lines = a.flatten.compact.prepend linex
+    lines.join("\n").lines\
+        .map {|line|  line =~ /.text$/ ? 'puts ' + line : line }.join
+
+  end
+
+  private
+
+  def formatline(pid, eid=nil, key=nil, tail=nil, index: nil)
+
+    if eid then
+
+      nametip = @nametip && tail.is_a?(String)
+      klass = nametip ? key.scan(/@class=['"]([^'"]+)/).last : nil
+
+      line = if klass then
+        desc = klass[0][/^[^\-]+/].gsub(/(?=[A-Z])/,' ').downcase
+        desc += " (e.g. %s)" % [tail.length < 50 ? tail : tail[0..46] + '...']
+        "\n# " + desc + "\n"
+      else
+        ''
+      end
+
+      line += "%s = %s.element(\"%s\")" % [eid, pid, key]
+      if tail.is_a? String
+        line += '.text'
+        #line += "\n" if nametip
+      end
+
+    else
+      line = index ? ("%s[%d].text" % [pid, index]) : ("%s.text" % pid)
+    end
+
+    return line
+  end
+
+
+  def getid(rawtag)
+
+    rawtagx = rawtag.split('/').last[/\w+/]
+
+    tag = case rawtagx.to_sym
+    when :a
+      'link'
+    else
+      rawtagx
+    end
+
+    if @tags.include?(tag) then
+      @tags[tag] =~ /\d+$/ ? @tags[tag].succ! : @tags[tag] += '1'
+    else
+      @tags[tag] = tag
+    end
+
+  end
+
+
   # Groups xpath by matching branches
   #
   def group_by_xpath(a)
@@ -48,39 +117,39 @@ class XMLColFinder
 
     h2 = {}
 
-    n = 0
-    while n < a.length
+    a.each do |path,txt|
 
-      path, txt = a[n]
       stickypath = ''
 
-      path.each do |name|
+      n = 0
+      while n < path.length
 
-        if @debug then
-          puts "h[+ stickypath + '/' + name]: " \
-              + h[stickypath + '/' + name].inspect
-        end
+        name = path[n]
 
-        if h[stickypath + '/' + name] > 1 then
+        if h[stickypath + '/' + name] > 1
 
           stickypath += '/' + name
-          next
+
+          if (n == path.length - 1) then
+
+            h2[stickypath.sub(/^\//,'')] ||= []
+            h2[stickypath.sub(/^\//,'')] << txt
+
+          end
 
         else
 
-          h2[stickypath] ||= []
-          h2[stickypath] << [path.join('/'), txt]
+
+          h2[stickypath.sub(/^\//,'')] ||= []
+          h2[stickypath.sub(/^\//,'')] << [path.join('/'), txt]
           break
 
         end
-
+        n += 1
       end
 
-      n += 1
-
     end
 
-
     if h2.length > 1 then
 
       a2 = h2.map {|k,v| [k.split('/'), v]}
@@ -91,32 +160,107 @@ class XMLColFinder
       return h2
 
     end
+
   end
 
-  def truncate_xpath(records, offset=0)
+  def scan(a, eid='doc', pid=eid.clone)
+
+    #puts 'a: ' + a.inspect if @debug
+
+    a.map do |row|
+
+      head, tail = row
+
+      if head.is_a? Array then
 
-    records.map do |key, value|
+        hline = scan(row, eid, pid)
 
-      new_key = key.sub(/^\/+/,'')[offset..-1]
-      len = new_key.length
-      #puts len.inspect
-      puts 'new_key: ' + new_key.inspect if @debug
+      elsif head
 
-      new_value = value.map do |k2, v2|
+        if head[0] == '/' then
 
-        puts 'k2: ' + k2.inspect if @debug
-        new_k2 = k2.sub(/^\/+/,'')[offset+len..-1]
-        new_len = new_k2.length
-        puts 'new_k2: ' + new_k2.inspect if @debug
-        puts 'v2: ' + v2.inspect if @debug
+          key = head[1..-1]
+          puts 'key: ' + key.inspect if @debug
 
-        v2b = v2.is_a?(Array) ? truncate_xpath(v2, offset+len+new_len) : v2
+          eid = getid(key)
 
-        [new_k2, v2b]
+          hline = if tail.is_a? Array and tail.all? {|x| x.is_a? String } then
+            @prev_xpath = true
+            "%s = %s.xpath(\"%s\")" % [eid, pid, key]
+          else
+            @prev_xpath = false
+            formatline(pid, eid, key, tail)
+          end
 
+        else
+
+          hline = formatline(pid=eid)
+        end
       end
 
-      [new_key, new_value]
+      if tail.is_a? Array then
+
+        if tail.compact[0].is_a? Array then
+
+          puts 'tail: ' + tail.inspect if @debug
+
+          tline = scan(tail, eid)
+
+        elsif tail.all? {|x| x.is_a? String} and tail[0][0] != '/'
+          puts '_tail: ' + tail.inspect if @debug
+          tline = tail.map.with_index do |x,i|
+            formatline(pid=eid, index: i)
+          end
+
+        end
+      end
+
+      [hline, tline]
+    end
+
+  end
+
+
+  def truncate_xpath(records, offset=0)
+
+    records.map do |record|
+
+      next unless record
+      if record.is_a? Array and (record[0].is_a? Array or record.length > 2) then
+
+        truncate_xpath(record, offset)
+
+      else
+
+        if record.is_a? String then
+
+          value = record
+          puts 'valuex: ' + value.inspect if @debug
+          value
+
+        else
+
+          key, value = record
+          puts 'key: ' + key.inspect if @debug
+
+          new_key = key[offset..-1]
+          puts 'new_key: ' + new_key.inspect if @debug
+
+          next unless new_key
+          len = new_key.length
+
+          if value.is_a? Array then
+
+            [new_key, truncate_xpath(value, offset+len)]
+
+          else
+
+            puts 'value: ' + value.inspect if @debug
+            [new_key, value]
+
+          end
+        end
+      end
 
     end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: xml_col_finder
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.2
 platform: ruby
 authors:
 - James Robertson
@@ -35,7 +35,7 @@ cert_chain:
   GSlZ9ilAfm8srTjbZ2cWQyNGGxH+zHQ3Z02c4ZEtgPv/wHjptd1VeBm0P1aemsRA
   ShsxXxzmzIrRENmpBp3tyR3k
   -----END CERTIFICATE-----
-date: 2022-01-22 00:00:00.000000000 Z
+date: 2022-01-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rexle